7 Man Syzygy and SSD

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
MikeB
Posts: 3391
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

7 Man Syzygy and SSD

Post by MikeB » Tue Dec 18, 2018 3:53 pm

The purpose of this post is to help one decide which 7 man syzygy files , along with the nearly 200GB files needed for 6 Man Syzygy to place on the a 1TB SSD drive to get maximum benefit.

I'm just going to put this out there as sort of help post for those that are wondering about the most economical way to use 7 man syzygy. I'm sure others have some pretty good suggestions as well and please feel free to share. First, there is real no need to get all 20 GB or so to benefit from 7 man syzygy. Fromm a practical standpoint , about 6o files (rtbw and rtbz) are all you need to get started , provided you have the all of the 6 man syzygy. Only RTBW are used during search, but the RTBZ files are needed, and all the tables that can be reached through promotion and the corresponding DTZ tables to play out the mate reliably. For this purpose, it assumed you will place the others files required to play out mate on an HDD drive.

So we have an 1 TB SSD drive, we have copied all of the 6 man rtbw and rtbz (recommended) onto this drive and we have roughly 800 GB left. A portfolio of the 7 man rtbw files listed below, will capture nearly 80% of the 7 man endings reached in human play.

Code: Select all

Size	EGTB		Bytes
--------------------------------------
    77G	 KBNPvKRP.rtbw	82,917,193,040
    75G	 KRBPvKRB.rtbw	80,280,358,288
    69G	 KRNPvKRN.rtbw	74,082,453,136
    68G	 KRNPvKRB.rtbw	72,853,318,096
    54G	 KRNPvKRP.rtbw	58,025,199,248
    45G	 KRBPvKRP.rtbw	48,381,043,216
    39G	 KRNPvKBN.rtbw	42,185,842,192
    29G	 KBPPvKRP.rtbw	31,606,770,192
    29G	 KBPPvKBP.rtbw	30,746,474,768
    28G	 KNPPvKRP.rtbw	29,911,219,856
    27G	 KNPPvKNP.rtbw	29,087,503,696
    27G	 KBPPvKNP.rtbw	28,727,898,832
    27G	 KRPPvKRB.rtbw	28,687,360,336
    26G	 KRPPvKRN.rtbw	27,834,688,208
    22G	 KRPPvKRP.rtbw	23,261,437,520
    21G	 KQPPvKQP.rtbw	22,632,171,664
    16G	 KRPPvKQP.rtbw	16,689,932,048
    11G	 KRPPvKBP.rtbw	11,435,368,336
   9.0G	 KNPPvKPP.rtbw	9,689,699,280
   9.0G	 KRPPvKNP.rtbw	9,632,432,848
   7.4G	 KBPPvKPP.rtbw	7,973,639,248
   7.1G	 KPPPvKBP.rtbw	7,584,064,976
   6.2G	 KPPPvKNP.rtbw	6,680,648,720
   4.7G	 KRPPvKPP.rtbw	5,054,627,664
   4.3G	 KPPPvKRP.rtbw	4,645,229,072
   2.1G	 KPPPvKQP.rtbw	2,269,738,128
   2.0G	 KPPPvKPP.rtbw	2,107,379,792
   1.0G	 KRPPPvKR.rtbw	1,075,778,832
   686M	 KRNPvKQP.rtbw	719,440,845
   ------------------------------------
			796,778,912,077
Happy Endgame Analysis!

User avatar
Nordlandia
Posts: 2406
Joined: Fri Sep 25, 2015 7:38 pm
Location: Sortland, Norway

Re: 7 Man Syzygy and SSD

Post by Nordlandia » Tue Dec 18, 2018 7:31 pm

If i'm not wrong 6-man rtbz can safely be stored on HDD. So don't waste essential RTBW space.

I'll recommend:

#1 | KRPPKRP

#2 | KQPPKQP

#3 | KNPPKNP

MikeB
Posts: 3391
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: 7 Man Syzygy and SSD

Post by MikeB » Tue Dec 18, 2018 9:44 pm

Nordlandia wrote:
Tue Dec 18, 2018 7:31 pm
If i'm not wrong 6-man rtbz can safely be stored on HDD. So don't waste essential RTBW space.

I'll recommend:

#1 | KRPPKRP

#2 | KQPPKQP

#3 | KNPPKNP
Yes, of course. It is, however, not one size fits all. As an example I always engine test using 6 man syzygy, so I have good reason to prefer having 6 man RTBZ on SSD. No right or wrong answer - it depends on one's usage, if one only use their engine set up for analysis, your suggestion is a good one. Thanks for sharing.

kasinp
Posts: 188
Joined: Sat Dec 02, 2006 9:47 pm
Location: Toronto

Re: 7 Man Syzygy and SSD

Post by kasinp » Wed Dec 19, 2018 12:21 am

MikeB wrote:
Tue Dec 18, 2018 9:44 pm
Nordlandia wrote:
Tue Dec 18, 2018 7:31 pm
If i'm not wrong 6-man rtbz can safely be stored on HDD. So don't waste essential RTBW space.

I'll recommend:

#1 | KRPPKRP

#2 | KQPPKQP

#3 | KNPPKNP
Yes, of course. It is, however, not one size fits all. As an example I always engine test using 6 man syzygy, so I have good reason to prefer having 6 man RTBZ on SSD. No right or wrong answer - it depends on one's usage, if one only use their engine set up for analysis, your suggestion is a good one. Thanks for sharing.
I decided to test the impact of having DTZ on HDD vs. SSD and concluded that their placement made no difference to the outcome of a series of 1+3 test games. I did it because like you (I think) I was sceptical. Now I am convinced :).
PK

duncan
Posts: 10202
Joined: Mon Jul 07, 2008 8:50 pm

Re: 7 Man Syzygy and SSD

Post by duncan » Wed Dec 19, 2018 12:54 am

MikeB wrote:
Tue Dec 18, 2018 3:53 pm
The purpose of this post is to help one decide which 7 man syzygy files , along with the nearly 200GB files needed for 6 Man Syzygy to place on the a 1TB SSD drive to get maximum benefit.

I'm just going to put this out there as sort of help post for those that are wondering about the most economical way to use 7 man syzygy. I'm sure others have some pretty good suggestions as well and please feel free to share. First, there is real no need to get all 20 GB or so to benefit from 7 man syzygy. Fromm a practical standpoint , about 6o files (rtbw and rtbz) are all you need to get started , provided you have the all of the 6 man syzygy. Only RTBW are used during search, but the RTBZ files are needed, and all the tables that can be reached through promotion and the corresponding DTZ tables to play out the mate reliably. For this purpose, it assumed you will place the others files required to play out mate on an HDD drive.

So we have an 1 TB SSD drive, we have copied all of the 6 man rtbw and rtbz (recommended) onto this drive and we have roughly 800 GB left. A portfolio of the 7 man rtbw files listed below, will capture nearly 80% of the 7 man endings reached in human play.

Code: Select all

Size	EGTB		Bytes
--------------------------------------
    77G	 KBNPvKRP.rtbw	82,917,193,040
    75G	 KRBPvKRB.rtbw	80,280,358,288
    69G	 KRNPvKRN.rtbw	74,082,453,136
    68G	 KRNPvKRB.rtbw	72,853,318,096
    54G	 KRNPvKRP.rtbw	58,025,199,248
    45G	 KRBPvKRP.rtbw	48,381,043,216
    39G	 KRNPvKBN.rtbw	42,185,842,192
    29G	 KBPPvKRP.rtbw	31,606,770,192
    29G	 KBPPvKBP.rtbw	30,746,474,768
    28G	 KNPPvKRP.rtbw	29,911,219,856
    27G	 KNPPvKNP.rtbw	29,087,503,696
    27G	 KBPPvKNP.rtbw	28,727,898,832
    27G	 KRPPvKRB.rtbw	28,687,360,336
    26G	 KRPPvKRN.rtbw	27,834,688,208
    22G	 KRPPvKRP.rtbw	23,261,437,520
    21G	 KQPPvKQP.rtbw	22,632,171,664
    16G	 KRPPvKQP.rtbw	16,689,932,048
    11G	 KRPPvKBP.rtbw	11,435,368,336
   9.0G	 KNPPvKPP.rtbw	9,689,699,280
   9.0G	 KRPPvKNP.rtbw	9,632,432,848
   7.4G	 KBPPvKPP.rtbw	7,973,639,248
   7.1G	 KPPPvKBP.rtbw	7,584,064,976
   6.2G	 KPPPvKNP.rtbw	6,680,648,720
   4.7G	 KRPPvKPP.rtbw	5,054,627,664
   4.3G	 KPPPvKRP.rtbw	4,645,229,072
   2.1G	 KPPPvKQP.rtbw	2,269,738,128
   2.0G	 KPPPvKPP.rtbw	2,107,379,792
   1.0G	 KRPPPvKR.rtbw	1,075,778,832
   686M	 KRNPvKQP.rtbw	719,440,845
   ------------------------------------
			796,778,912,077
Happy Endgame Analysis!
Would you know the elo gain over if you had those 7 man over 6 man on your set up?

jdart
Posts: 3816
Joined: Fri Mar 10, 2006 4:23 am
Location: http://www.arasanchess.org

Re: 7 Man Syzygy and SSD

Post by jdart » Wed Dec 19, 2018 1:02 am

Would you know the elo gain over if you had those 7 man over 6 man on your set up?
I think the gain even from 6-man is quite small, because it is seldom you go into an ending that can't be solved by search or handled by special case evals. Those endings certainly exist, the question is how often do they occur in real games? I think not often.

--Jon

glennsamuel32
Posts: 85
Joined: Sat Dec 04, 2010 4:31 pm
Location: 223

Re: 7 Man Syzygy and SSD

Post by glennsamuel32 » Wed Dec 19, 2018 5:54 am


MikeB
Posts: 3391
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: 7 Man Syzygy and SSD

Post by MikeB » Wed Dec 19, 2018 6:18 am

duncan wrote:
Wed Dec 19, 2018 12:54 am
MikeB wrote:
Tue Dec 18, 2018 3:53 pm
The purpose of this post is to help one decide which 7 man syzygy files , along with the nearly 200GB files needed for 6 Man Syzygy to place on the a 1TB SSD drive to get maximum benefit.

I'm just going to put this out there as sort of help post for those that are wondering about the most economical way to use 7 man syzygy. I'm sure others have some pretty good suggestions as well and please feel free to share. First, there is real no need to get all 20 GB or so to benefit from 7 man syzygy. Fromm a practical standpoint , about 6o files (rtbw and rtbz) are all you need to get started , provided you have the all of the 6 man syzygy. Only RTBW are used during search, but the RTBZ files are needed, and all the tables that can be reached through promotion and the corresponding DTZ tables to play out the mate reliably. For this purpose, it assumed you will place the others files required to play out mate on an HDD drive.

So we have an 1 TB SSD drive, we have copied all of the 6 man rtbw and rtbz (recommended) onto this drive and we have roughly 800 GB left. A portfolio of the 7 man rtbw files listed below, will capture nearly 80% of the 7 man endings reached in human play.

Code: Select all

Size	EGTB		Bytes
--------------------------------------
    77G	 KBNPvKRP.rtbw	82,917,193,040
    75G	 KRBPvKRB.rtbw	80,280,358,288
    69G	 KRNPvKRN.rtbw	74,082,453,136
    68G	 KRNPvKRB.rtbw	72,853,318,096
    54G	 KRNPvKRP.rtbw	58,025,199,248
    45G	 KRBPvKRP.rtbw	48,381,043,216
    39G	 KRNPvKBN.rtbw	42,185,842,192
    29G	 KBPPvKRP.rtbw	31,606,770,192
    29G	 KBPPvKBP.rtbw	30,746,474,768
    28G	 KNPPvKRP.rtbw	29,911,219,856
    27G	 KNPPvKNP.rtbw	29,087,503,696
    27G	 KBPPvKNP.rtbw	28,727,898,832
    27G	 KRPPvKRB.rtbw	28,687,360,336
    26G	 KRPPvKRN.rtbw	27,834,688,208
    22G	 KRPPvKRP.rtbw	23,261,437,520
    21G	 KQPPvKQP.rtbw	22,632,171,664
    16G	 KRPPvKQP.rtbw	16,689,932,048
    11G	 KRPPvKBP.rtbw	11,435,368,336
   9.0G	 KNPPvKPP.rtbw	9,689,699,280
   9.0G	 KRPPvKNP.rtbw	9,632,432,848
   7.4G	 KBPPvKPP.rtbw	7,973,639,248
   7.1G	 KPPPvKBP.rtbw	7,584,064,976
   6.2G	 KPPPvKNP.rtbw	6,680,648,720
   4.7G	 KRPPvKPP.rtbw	5,054,627,664
   4.3G	 KPPPvKRP.rtbw	4,645,229,072
   2.1G	 KPPPvKQP.rtbw	2,269,738,128
   2.0G	 KPPPvKPP.rtbw	2,107,379,792
   1.0G	 KRPPPvKR.rtbw	1,075,778,832
   686M	 KRNPvKQP.rtbw	719,440,845
   ------------------------------------
			796,778,912,077
Happy Endgame Analysis!
Would you know the elo gain over if you had those 7 man over 6 man on your set up?
The setup described above with 30 7 man EGTB on SSD and the balance on HDD is probably in the range of 1 to 2 Elo at fast time controls ( bullet like) , over 6 man EGTB, less Elo gain at longer time controls. Going back to the 4 man EGTB, it probably gain 5 or 6 Elo compared to none, going from 4 to 5 Man EGTB , perhaps gain 2 or 3 Elo, going from 5 man to 6 man EGTB maybe gain another 1-2 Elo and like wise with going from 6 to 7 man EGTB, another 1-2 Elo. And alway less at longer TC.

Just for kicks, ran identical engines with McCain-X below accessing the complete 7 man set and with the identical engine McCain-GH accessing only the complete 6 man EGTB. This was micro bullet tc of 15 sec game with 0.5 second increment.

Code: Select all

ResultSet>rp /Users/michaelbyrne/cluster.mfb/12192018.pgn
400 game(s) loaded
ResultSet>elo
ResultSet-EloRating>mm 0 1
Iteration 100: 1.26704e-05 
00:00:00,00
ResultSet-EloRating>r
Rank Name       Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR 
---------------------------------------------------------------------------------------------------------
   1 McCain X    3101   0.0   14   14   400  201.0  50.2   37   35  328   9.2  82.0  3099 
   2 McCain GH   3099   1.6   14   14   400  199.0  49.8   35   37  328   8.8  82.0  3101 
---------------------------------------------------------------------------------------------------------
Not statistically valid by any means, but it does indicate there may be a slight 1 to 2 Elo gain from 6 to 7 man EGTB ( using the described setup above) at very fast tc. ymmv.

duncan
Posts: 10202
Joined: Mon Jul 07, 2008 8:50 pm

Re: 7 Man Syzygy and SSD

Post by duncan » Wed Dec 19, 2018 9:44 am

MikeB wrote:
Wed Dec 19, 2018 6:18 am
duncan wrote:
Wed Dec 19, 2018 12:54 am
MikeB wrote:
Tue Dec 18, 2018 3:53 pm
The purpose of this post is to help one decide which 7 man syzygy files , along with the nearly 200GB files needed for 6 Man Syzygy to place on the a 1TB SSD drive to get maximum benefit.

I'm just going to put this out there as sort of help post for those that are wondering about the most economical way to use 7 man syzygy. I'm sure others have some pretty good suggestions as well and please feel free to share. First, there is real no need to get all 20 GB or so to benefit from 7 man syzygy. Fromm a practical standpoint , about 6o files (rtbw and rtbz) are all you need to get started , provided you have the all of the 6 man syzygy. Only RTBW are used during search, but the RTBZ files are needed, and all the tables that can be reached through promotion and the corresponding DTZ tables to play out the mate reliably. For this purpose, it assumed you will place the others files required to play out mate on an HDD drive.

So we have an 1 TB SSD drive, we have copied all of the 6 man rtbw and rtbz (recommended) onto this drive and we have roughly 800 GB left. A portfolio of the 7 man rtbw files listed below, will capture nearly 80% of the 7 man endings reached in human play.

Code: Select all

Size	EGTB		Bytes
--------------------------------------
    77G	 KBNPvKRP.rtbw	82,917,193,040
    75G	 KRBPvKRB.rtbw	80,280,358,288
    69G	 KRNPvKRN.rtbw	74,082,453,136
    68G	 KRNPvKRB.rtbw	72,853,318,096
    54G	 KRNPvKRP.rtbw	58,025,199,248
    45G	 KRBPvKRP.rtbw	48,381,043,216
    39G	 KRNPvKBN.rtbw	42,185,842,192
    29G	 KBPPvKRP.rtbw	31,606,770,192
    29G	 KBPPvKBP.rtbw	30,746,474,768
    28G	 KNPPvKRP.rtbw	29,911,219,856
    27G	 KNPPvKNP.rtbw	29,087,503,696
    27G	 KBPPvKNP.rtbw	28,727,898,832
    27G	 KRPPvKRB.rtbw	28,687,360,336
    26G	 KRPPvKRN.rtbw	27,834,688,208
    22G	 KRPPvKRP.rtbw	23,261,437,520
    21G	 KQPPvKQP.rtbw	22,632,171,664
    16G	 KRPPvKQP.rtbw	16,689,932,048
    11G	 KRPPvKBP.rtbw	11,435,368,336
   9.0G	 KNPPvKPP.rtbw	9,689,699,280
   9.0G	 KRPPvKNP.rtbw	9,632,432,848
   7.4G	 KBPPvKPP.rtbw	7,973,639,248
   7.1G	 KPPPvKBP.rtbw	7,584,064,976
   6.2G	 KPPPvKNP.rtbw	6,680,648,720
   4.7G	 KRPPvKPP.rtbw	5,054,627,664
   4.3G	 KPPPvKRP.rtbw	4,645,229,072
   2.1G	 KPPPvKQP.rtbw	2,269,738,128
   2.0G	 KPPPvKPP.rtbw	2,107,379,792
   1.0G	 KRPPPvKR.rtbw	1,075,778,832
   686M	 KRNPvKQP.rtbw	719,440,845
   ------------------------------------
			796,778,912,077
Happy Endgame Analysis!
Would you know the elo gain over if you had those 7 man over 6 man on your set up?
The setup described above with 30 7 man EGTB on SSD and the balance on HDD is probably in the range of 1 to 2 Elo at fast time controls ( bullet like) , over 6 man EGTB, less Elo gain at longer time controls. Going back to the 4 man EGTB, it probably gain 5 or 6 Elo compared to none, going from 4 to 5 Man EGTB , perhaps gain 2 or 3 Elo, going from 5 man to 6 man EGTB maybe gain another 1-2 Elo and like wise with going from 6 to 7 man EGTB, another 1-2 Elo. And alway less at longer TC.

Just for kicks, ran identical engines with McCain-X below accessing the complete 7 man set and with the identical engine McCain-GH accessing only the complete 6 man EGTB. This was micro bullet tc of 15 sec game with 0.5 second increment.

Code: Select all

ResultSet>rp /Users/michaelbyrne/cluster.mfb/12192018.pgn
400 game(s) loaded
ResultSet>elo
ResultSet-EloRating>mm 0 1
Iteration 100: 1.26704e-05 
00:00:00,00
ResultSet-EloRating>r
Rank Name       Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR 
---------------------------------------------------------------------------------------------------------
   1 McCain X    3101   0.0   14   14   400  201.0  50.2   37   35  328   9.2  82.0  3099 
   2 McCain GH   3099   1.6   14   14   400  199.0  49.8   35   37  328   8.8  82.0  3101 
---------------------------------------------------------------------------------------------------------
Not statistically valid by any means, but it does indicate there may be a slight 1 to 2 Elo gain from 6 to 7 man EGTB ( using the described setup above) at very fast tc. ymmv.
Thanks for the experiment. Could you post the 2 games McCain-X won due to the complete 7 man set and do you know at what time scale it would have drawn ?

MikeB
Posts: 3391
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: 7 Man Syzygy and SSD

Post by MikeB » Wed Dec 19, 2018 12:18 pm

duncan wrote:
Wed Dec 19, 2018 9:44 am

<snip>
Thanks for the experiment. Could you post the 2 games McCain-X won due to the complete 7 man set and do you know at what time scale it would have drawn ?
I will give it a shot. Since I have doing this for nearly 40 years, I have looked at games for this type of action many times in the past. When EGTB first came out, you would be able to find quite a few games (out of thousands) that were won as the result of EGTB. But that was comparing EGTB to no EGTB. When you compare 7 man EGTB to 6 man EGTB, and if you are using the 50 move draw rule, you might really be looking for game that was "lost" and turned into a draw. You got me thinking about running the same test with 50 move rule turned off. I will skim through the 400 games played here and then rerun the 400 game set with 50 move rule turned off and see what happens there. Will post what I find later this AM ( currently 7:15 AM here).

Post Reply